What is Metered API Access?
Metered API access is a billing model where API consumers pay based on actual usage — per request, per token generated, per byte transferred, or per computation performed — rather than prepaid credits or flat subscriptions.
WHY IT MATTERS
Traditional API billing models were designed for human developers with predictable usage patterns: monthly subscriptions, prepaid credit bundles, or tiered plans. These models break down when AI agents consume APIs with wildly variable demand — an agent might make zero calls for days, then burst to thousands of calls in minutes.
The x402 protocol enables true metered access at the HTTP level. With the exact scheme, each request costs a fixed amount — no accounts, no prepaid balance, no overage charges. The planned upto scheme takes this further by allowing dynamic metering: the client authorises up to a maximum amount, and the server settles based on actual consumption (e.g. number of tokens generated by an LLM).
Metered access eliminates three problems simultaneously: overprovisioning (buying credits you don't use), access friction (signing up, KYC, payment method setup), and vendor lock-in (switching providers means losing prepaid credits). Agents can freely switch between providers based on quality, latency, and real-time pricing.
The economic model works because blockchain settlement costs are negligible on L2s like Base — sub-cent transaction fees make it viable to charge fractions of a cent per API call.
HOW POLICYLAYER USES THIS
PolicyLayer tracks metered API spending with aggregate daily caps and per-endpoint budgets. Even when individual requests are cheap, high-frequency agent consumption can accumulate rapidly. PolicyLayer's rate limiting and daily spending limits prevent metered access from becoming an unbounded cost.